Efficacy and safety of anti-hyperglycaemic drugs in patients with non-alcoholic fatty liver disease with or without diabetes: An updated systematic review of randomized controlled trials

Journal Pre-proof Efficacy and safety of anti-hyperglycaemic drugs in patients with non-alcoholic fatty liver disease with or without diabetes: an updated systematic review of randomized controlled trials Alessandro Mantovani MD Christopher D. Byrne MB BCh PhD Eleonora Scorletti MD Christos S. Mantzoros MD Giovanni Targher MD

PII:

S1262-3636(20)30002-1

DOI:

https://doi.org/doi:10.1016/j.diabet.2019.12.007

Reference:

DIABET 101142

To appear in:

Diabetes & Metabolism

Received Date:

21 November 2019

Revised Date:

26 December 2019

Accepted Date:

27 December 2019

Please cite this article as: Mantovani A, Byrne CD, Scorletti E, Mantzoros CS, Targher G, Efficacy and safety of anti-hyperglycaemic drugs in patients with non-alcoholic fatty liver disease with or without diabetes: an updated systematic review of randomized controlled trials, Diabetes and Metabolism (2020), doi: https://doi.org/10.1016/j.diabet.2019.12.007

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.

Efficacy and safety of anti-hyperglycaemic drugs in patients with nonalcoholic fatty liver disease with or without diabetes: an updated systematic review of randomized controlled trials Short title: Anti-hyperglycaemic drugs for NAFLD treatment Alessandro Mantovani, MD1*, Christopher D. Byrne, MB BCh, PhD2,3*, Eleonora Scorletti, MD2,3,4, Christos S. Mantzoros, MD5, Giovanni Targher, MD1 1

of

Section of Endocrinology, Diabetes and Metabolism, Department of Medicine, University and Azienda Ospedaliera Universitaria Integrata of Verona, Verona, Italy 2

ro

Southampton National Institute for Health Research Biomedical Research Centre, University Hospital Southampton, Southampton General Hospital, Southampton, UK 3

-p

Nutrition and Metabolism, Faculty of Medicine, University of Southampton, UK

4

Division of Gastroenterology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

5

re

Division of Endocrinology, Diabetes and Metabolism, Department of Internal Medicine, Beth Israel Deaconess Medical Centre, Harvard Medical School, Boston, MA, USA

rn al P

*These authors have contributed equally to the manuscript. Address for correspondence: Prof. Giovanni Targher

Section of Endocrinology, Diabetes and Metabolism Department of Medicine

University and Azienda Ospedaliera Universitaria Integrata

Jo u

Piazzale Stefani, 1

37126 Verona, Italy

Phone: +39/045-8123110

E-mail: [email protected]

1

Page 1 of 38

ABSTRACT Aim. - There are no approved drugs for the treatment of non-alcoholic fatty liver disease (NAFLD). However, many randomized controlled trials (RCT) have examined the effect of anti-hyperglycaemic agents on NAFLD in patients with and without type 2 diabetes mellitus (T2DM), since both T2DM and insulin resistance are closely linked to this burdensome liver disease. Methods. - We systematically searched publication databases using predefined keywords to identify headto-head or placebo-controlled RCTs (published until September 30, 2019) of NAFLD individuals testing the

of

efficacy of anti-hyperglycaemic drugs to specifically treat NAFLD or non-alcoholic steatohepatitis (NASH).

ro

Outcomes of interest included changes in serum aminotransferase levels, liver fat, liver fibrosis, or histologic resolution of NASH.

-p

Results. - We included 29 RCTs involving a total of 2,617 individuals (~45% had T2DM) that have used

re

metformin (n=6 studies), glitazones (n=8 studies), glucagon-like peptide-1 receptor agonists (n=6 studies), dipeptidyl peptidase-4 inhibitors (n=4 studies) or sodium-glucose cotransporter-2 inhibitors (n=7 studies) to

rn al P

treat NAFLD. Although most anti-hyperglycaemic drugs improved serum liver enzymes, only glitazones (especially pioglitazone) and liraglutide showed an improvement of histologic features of NAFLD, with a mild beneficial effect also on liver fibrosis for pioglitazone only. Conclusion. - RCT evidence supports the efficacy of some anti-hyperglycaemic agents (especially pioglitazone)

Jo u

in patients with NAFLD or NASH, though weight gain with pioglitazone may warrant caution. Further welldesigned RCTs are needed to better characterize the efficacy and safety of monotherapy and combination therapy with anti-hyperglycaemic agents in patients with NAFLD. Key-words: Anti-hyperglycaemic drugs; NAFLD; NASH; Type 2 diabetes INTRODUCTION Non-alcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide, affecting up to nearly 25-30% of adults in the general population, nearly 55-60% of patients with type 2 diabetes mellitus

2

Page 2 of 38

(T2DM) and the large majority of those with severe obesity. Worryingly, the prevalence of NAFLD is expected to rise further over the next decade [1,2]. The pathophysiology of NAFLD is complex and intricate, but NAFLD and T2DM form part of a vicious spiral of disease affecting both conditions. On the one hand, NAFLD is strongly associated with T2DM and obesity, i.e., two pathologic conditions that predict the development of non-alcoholic steatohepatitis (NASH), advanced fibrosis and cirrhosis, which are the histological features more consistently associated with adverse hepatic and extra-hepatic outcomes in NAFLD [1,2]. On the other hand, NAFLD is also associated with an

of

approximate twofold increased risk of incident T2DM [3,4].

ro

The management of NAFLD is essentially based on lifestyle modification(s) and early treatment of associated

-p

metabolic co-morbidities [5,6]. Although currently there are no approved drug treatments for NAFLD, seeing that T2DM is linked to NAFLD and its more advanced forms [1,2], an ever-increasing number of non-

re

randomized interventional studies and randomized controlled trials (RCTs) have focused on testing the

rn al P

efficacy of anti-hyperglycaemic agents in patients with NAFLD [7,8]. Such anti-hyperglycaemic agents include metformin, glitazones, glucagon-like peptide-1 receptor agonists (GLP-1RAs), dipeptidyl peptidase 4 (DPP-4) inhibitors and sodium-glucose cotransporter 2 (SGLT-2) inhibitors. Therefore, our aim was to undertake an updated systematic review of published RCTs that have evaluated

NASH.

Jo u

the efficacy and safety of the aforementioned anti-hyperglycaemic agents to specifically treat NAFLD or

MATERIALS AND METHODS

Search strategy and study selection We followed systematic review methodology and procedures that are in accordance with current guidance for systematic reviews [i.e., Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA); http://www.prisma-statement.org/] [9].

3

Page 3 of 38

We included head-to-head or placebo-controlled RCTs of adults or children with NAFLD, which used a European Medicines Agency (EMA)-approved anti-hyperglycaemic drug for treatment of NAFLD or NASH. We included only RCTs that had at least 20 patients per treatment arms of interest. Relevant studies were identified by systematically searching PubMed, ClinicalTrials.Gov and Cochrane Database of Systematic Reviews from January 1, 1990 to 30 September, 2019 (date last searched) using the free text terms ‘non-alcoholic fatty liver disease’ (OR ‘NAFLD’ OR ‘non-alcoholic steatohepatitis’ OR ‘NASH’) AND ‘metformin’ OR ‘thiazolidinediones’ OR ‘glitazones’ OR ‘glucagon-like peptide-1 receptor agonists’ OR

of

‘dipeptidyl peptidase-4 inhibitors’ OR ‘sodium-glucose cotransporter-2 inhibitors’. Searches were restricted

ro

to head-to-head or placebo-controlled RCTs where the diagnosis of NAFLD was based on liver biopsy or imaging techniques. Reference lists of relevant papers and previous review articles were hand searched for

-p

other relevant studies. Studies reported only in conference abstracts, unpublished studies, retrospective

re

observational studies and non-randomized interventional studies were excluded. Studies in languages other than English were also excluded. Placebo-controlled RCTs examining the efficacy of sulphonylureas, acarbose

literature.

rn al P

or insulin on NASH resolution, liver fat content and other liver function parameters were not available in the

One investigator screened citations and a second investigator assessed excluded citations. Two investigators independently evaluated full-text articles by applying the inclusion criteria and resolved disagreements by

Jo u

consensus.

Data extraction and quality assessment For all RCTs, we extracted information on participants’ characteristics, interventions, methods used for diagnosing/staging NAFLD, and results for effectiveness and harms outcomes. In particular, the primary outcomes of interest included changes in serum alanine-aminotransferase (ALT) and aspartateaminotransferase (AST) levels, liver fat content, or histologic resolution of NASH and changes in individual histologic scores of NASH (i.e., steatosis, necro-inflammation and fibrosis). We also extracted information on

4

Page 4 of 38

weight loss, changes in haemoglobin A1c and serious adverse events, as well as percentage of withdrawals due to adverse events. Each RCT was assessed for quality by two independent reviewers, with disagreements resolved through consensus. Study quality was assessed according to predefined criteria that are based on those used by the US Preventive Services Task Force (USPSTF) and the National Health Service Centre for Reviews and Dissemination [9,10]. Specifically, these criteria focus on methods of randomization, allocation concealment, blinding of providers, outcome assessors, and patients; similarity of group characteristics at baseline;

of

attrition rate; and use of intention-to-treat analysis. RCTs that met all criteria were rated as good quality;

ro

RCTs with an element at high risk of bias or failed to meet combinations of criteria were rated as poor quality;

-p

and the remainder were rated as fair quality. RESULTS

re

Figure S1 (see supplementary materials associated with this article on line) displays the flow diagram of the

rn al P

literature research and study selection. Overall, we included 29 placebo-controlled or active-controlled RCTs for a total of 2,617 individuals (45% of them had established T2DM), who were treated for a median period of 6 months (inter-quartile range [IQR]: 5-12 months). Baseline characteristics of the eligible RCTs on metformin (n=6), glitazones (n=8), GLP-1RAs (n=6), DPP-4 inhibitors (n=4) or SGLT-2 inhibitors (n=7) are summarized in Tables I to IV, respectively. The eleven RCTs excluded at the stage of eligibility according to

Jo u

the flow diagram are listed in Table S1 (see supplementary materials associated with this article on line). Most eligible RCTs were small (n <40 per treatment arm) and rated as fair quality, primarily due to unclear blinding, unclear allocation concealment, high attrition or lack of liver histology endpoints. All eligible RCTs in NAFLD patients with and without T2DM included treatment with an anti-hyperglycaemic drug in one or more treatment arms. Metformin We included a total of six placebo-controlled or active-controlled RCTs that used metformin to treat NAFLD (Table I) [11-16]. These studies enrolled 573 individuals, most of whom (> 90%) did not have diabetes (76%

5

Page 5 of 38

men; mean age 38 ± 15 years; BMI 30 ± 2.5 kg/m2, AST 55 ± 11 UI/L, ALT 86 ± 27 UI/L), who were treated for a median of 9 months (IQR 6-12 months). Of the eligible RCTs, one was conducted in obese children/adolescents with biopsy-proven NASH (the TONIC trial), whereas the remaining five RCTs involved adults with or without T2DM. Three of the eligible RCTs were conducted in Europe, two in Asia and one in United States. Among the four RCTs including patients with biopsy-proven NAFLD (except for the TONIC trial that failed to show any beneficial effect), metformin showed small beneficial effects on liver steatosis and inflammation, but no significant effects on liver fibrosis and resolution of NASH [11-14]. In the two remaining

of

RCTs involving patients with imaging-defined NAFLD, metformin showed a neutral effect on liver steatosis,

ro

when compared to placebo or reference therapy [15,16]. In most of these eligible RCTs, metformin showed a significant reduction of serum aminotransferase levels (especially serum ALT). The effect of metformin on

-p

body weight was neutral, whereas there was a substantial improvement of HbA1c levels (~0.8-1%). Metformin was generally well tolerated, although a study reported a higher withdrawal due to adverse

rn al P

Glitazones

re

effects (mostly gastrointestinal symptoms) in the metformin group when compared to placebo [12].

As shown in Table II, we included a total of eight placebo-controlled or active-controlled RCTs that used either pioglitazone (n=6) or rosiglitazone (n=2) to treat NAFLD [14,16-22]. These RCTs included 828 individuals, most of whom (~85%) did not have diabetes (57% men; mean age 47 ± 7 years; BMI 31 ± 3 kg/m2;

Jo u

AST 54 ± 8 UI/L; ALT 80 ± 15 UI/L), who were treated for a median of 12 months (IQR 6-14 months). Only one study was undertaken in patients with imaging-defined NAFLD [15], whereas all other RCTs included patients with biopsy-proven NAFLD. Four RCTs were conducted in United States, two in Europe and two in Asia. When compared to placebo or reference therapy, both pioglitazone and rosiglitazone significantly improved liver fat content and even NASH [15,17-23]. With regard to a possible improvement of liver fibrosis, glitazones were not superior to placebo or other active molecules, except for one RCT using pioglitazone 45 mg/day for 18 months in patients with biopsy-proven NASH and T2DM/prediabetes [23]. A significant reduction of serum aminotransferase levels was observed in most patients treated with glitazones, when compared to placebo or reference therapy. Glitazones had a similar adverse event profile to placebo or reference therapy, with

6

Page 6 of 38

the exception of a weight gain (e.g., ~2-3 kg at a dosage of 45 mg/day of pioglitazone) [23]. Withdrawals due to serious adverse effects were not increased in the glitazone group compared to either placebo or other active agents. GLP-1RAs We included a total of six placebo-controlled or active-controlled RCTs that used liraglutide (n=4) or exenatide (n=2) to treat NAFLD (Table III) [24-29]. These RCTs included 396 individuals, the large majority

of

(73%) of whom had diabetes (51% men; mean age 49 ± 5 years; BMI 32 ± 4 kg/m2; AST 47 ± 39 UI/L; ALT 65 ± 52 UI/L) and who were treated for a median of 6 months (IQR 5.5-7.0 months). Four RCTs were undertaken

-p

involved non-diabetic women with polycystic ovary syndrome.

ro

in patients with T2DM, one RCT was conducted in both patients with and without T2DM, whereas one RCT

Three RCTs were conducted in Europe and three in China. Only a small phase-2b RCT (i.e., the LEAN trial)

re

included patients with biopsy-proven NASH [25], whereas in the other five RCTs, NAFLD was diagnosed by

rn al P

imaging techniques (ultrasonography or magnetic resonance imaging). When compared to placebo or reference therapy, GLP-1RAs (especially liraglutide) improved liver fat and reduced serum aminotransferase levels in a dose-dependent way, as well as body weight (~3-5 kg) and HbA1c levels (~1-1.2%). Data on liver fibrosis were available only in the LEAN trial and liraglutide failed to produce any significant histological improvement in liver fibrosis compared to placebo [25]. GLP-1RAs were well tolerated and had a similar

Jo u

adverse event profile to placebo (or reference therapy), except for an increased frequency of gastrointestinal symptoms, such as loss of appetite, nausea, constipation or diarrhoea. These events tended to be transient and mild-to-moderate in severity across most of the included RCTs. DPP-4 inhibitors We included a total of four placebo-controlled or active-controlled RCTs that used either sitagliptin (n=3) or vildagliptin (n=1) to treat NAFLD (Table III) [29-32]. These RCTs included a total of 241 individuals with T2DM or pre-diabetes (68% men; mean age 56 ± 9 years; BMI 29 ± 3 kg/m 2; AST 31 ± 3 UI/L; ALT 37 ± 7 UI/L), who were treated for a median of 6 months (IQR 5.5-8.0 months). In these four RCTs, NAFLD was detected

7

Page 7 of 38

exclusively by imaging techniques. Two RCTs were undertaken in China, one in United States and another one in United Kingdom. When compared to placebo or reference therapy, vildagliptin had a marginal significant effect on liver fat, whereas sitagliptin did not. Given the absence of liver histological data, we are unable to comment on the effect of DPP-4 inhibitors on histological resolution of NASH or other histologic features of NAFLD. When compared to placebo or reference therapy, DPP-4 inhibitors showed a reduction of HbA1c (~0.5-0.8%), but a neutral effect on body weight and serum liver enzymes, except for vildagliptin that showed a reduction (~7 IU/L) in serum ALT levels. DPP-4 inhibitors were well tolerated with a similar

of

adverse event profile to placebo or reference therapy.

ro

SGLT-2 inhibitors

-p

We included a total of seven placebo-controlled or active-controlled RCTs that used empagliflozin (n=2), dapagliflozin (n=3), canagliflozin (n=1) or ipragliflozin (n=1) to treat NAFLD (Table IV) [33-39]. These RCTs

re

included only individuals with T2DM (n=579; 58% men; mean age 58±5 years; BMI 31 ± 2 kg/m2; AST 34 ± 10

rn al P

UI/L; ALT 43 ± 16 UI/L), who were treated for a median of ~6 months (IQR 5.5-6.5 months). In these seven RCTs the diagnosis of NAFLD was based on imaging techniques (ultrasonography, magnetic resonance imaging or FibroScan®). One RCT included an international cohort, three RCTs were performed in Asia, one in United States and two in Europe. When compared to placebo or reference therapy, empagliflozin, canagliflozin or ipragliflozin (not available in Europe) showed a small improvement of liver fat content, along

Jo u

with a reduction of body weight (~2-3 kg) and HbA1c (~0.8-1.0%). By contrast, in the RCT of Bolinder et al., despite the lowering of body weight and glucose parameters, a 24-week treatment with dapagliflozin 10 mg once daily did not show any significant reduction of liver fat content (on magnetic resonance imaging) when compared to placebo [34]. In all RCTs, SGLT-2 inhibitors showed a significant reduction of serum aminotransferase levels. SGLT-2 inhibitors were well tolerated with a similar adverse event profile to placebo or reference therapy, except for increased risk of genitourinary infections. DISCUSSION

8

Page 8 of 38

Compared with other narrative review articles published on this topic, our systematic review provides the most updated evidence of placebo-controlled or active-controlled RCTs that examined the efficacy and safety of anti-hyperglycaemic agents (including newer anti-hyperglycaemic drugs, i.e., the SGLT-2 inhibitors, DPP-4 inhibitors and GLP-1RAs) to specifically treat NAFLD or NASH. Our systematic review includes 29 head-tohead or placebo-controlled RCTs (published until September 30, 2019), involving a total of 2,617 NAFLD individuals with and without T2DM. The primary outcomes of interest included changes in serum aminotransferase levels, liver fat, or histological resolution of NASH and changes in individual histologic

of

scores of NASH. We did not attempt to pool these RCTs into a meta-analysis due to the highly variable

ro

mechanisms of actions of the anti-hyperglycaemic agents in this analysis, as well as the high heterogeneity of the interventions, populations included and the outcome measures.

-p

From our systematic review, it clearly emerges that the major issue in this field of research is the scarcity of

re

high-quality, adequately powered RCTs of sufficient duration that include clinically relevant hepatic endpoints (i.e., liver histologic data). Some concerns also remain about the long-term safety of the available

rn al P

drugs, necessitating thoughtful balancing for the clinician and the patient of the potential risks and benefits. Therefore, to date, drug treatment for NAFLD is best targeted at patients with biopsy-confirmed NASH, who are at the highest risk of progressive liver disease [5,6]. Specifically, while most of RCTs using metformin or glitazones have investigated the efficacy of these drugs

Jo u

on the histologic features of NAFLD or resolution of NASH in biopsied patients [11-14,17-23], almost all of the published RCTs testing the hepatic effects of the newer anti-hyperglycaemic agents did not have any adequate liver histological endpoints [15,16,24,26-39], with the only exception of the LEAN trial, which is a phase 2b placebo-controlled RCT, enrolling 52 UK patients with biopsy-proven NASH, who were randomly assigned to liraglutide 1.8 mg/day or placebo [25]. This is an important weakness to consider in the interpretation of the main results of these published RCTs. RCTs of pharmacologic treatments aimed at improving liver disease severity in NAFLD should always include patients with biopsy‐confirmed NASH or fibrosis, not the least because liver fibrosis is the histological feature most strongly associated with increased risk of adverse clinical outcomes [1,5,40]. Histological endpoints are still considered the best predictors of

9

Page 9 of 38

cirrhosis and other liver-related outcomes, although they can only be reliably tested in phase 2 and 3 RCTs (but not during post-marketing monitoring trials) [40-42]. To date, phase 3 RCTs require a histological endpoint of NASH resolution without fibrosis progression. Additionally, although magnetic resonance imagingestimated proton density fat fraction can accurately quantify changes in liver fat content, its efficacy for detecting NASH and advanced fibrosis is rather limited [40]. Moreover, although liver steatosis assessed by magnetic resonance imaging-estimated proton density fat fraction is increasingly used in the early phase trials examining drugs with anti-steatotic effects, the prognostic significance of a reduction in liver steatosis

of

remains unclear [40-42]. Based on these considerations, most RCTs included in our systematic review have

ro

obtained a fair quality according to the USPSTF criteria.

Metformin is a biguanide, glucose-lowering drug that represents the first-line choice for treatment of T2DM

-p

[43]. Our systematic review corroborates the conclusion that metformin, despite its beneficial effects on

re

serum liver enzymes, does not exert any beneficial effect on liver histology features, NAFLD activity score, or resolution of NASH in both adults and children with NAFLD [11-16]. This finding confirms the American and

rn al P

European practice guidelines that recommended against the use of metformin to specifically treat NAFLD or NASH [5,6]. That said, however, it is also important to underline that some evidence is now suggesting a possible hepato-protective role of metformin for reducing risk of cirrhosis and hepatocellular carcinoma [7,8].

Jo u

Rosiglitazone and pioglitazone (i.e., this latter being the only glitazone drug now available on the market in most European countries) are selective ligands of the peroxisome-proliferator-activated receptor (PPAR)gamma [43]. The PPAR-gamma receptor has three isoforms. The PPAR-gamma receptor-2 isoform is highly expressed in adipose tissue, acting to redistribute adipose tissue between intra-abdominal and subcutaneous adipose tissue by promoting accumulation of triglyceride in peripheral fat depots. PPAR-gamma is also expressed in Kupffer cells and PPAR-gamma has potent anti-inflammatory action to decrease nuclear factorkB mediated cytokine and chemokine production, while at the same time increasing adiponectin levels [7,8], suggesting plausible biological mechanisms by which pioglitazone can improve liver disease, at least in some patients with NASH.

10

Page 10 of 38

Our systematic review shows that pioglitazone use in patients with NASH had significant benefits in liver function, liver fat content and resolution of NASH both in patients with and without T2DM [15,17-23], though increases in body weight and lower-limb oedema may be cause for concern. Evidence for rosiglitazone was more limited and had somewhat mixed results, but results were generally comparable to those for pioglitazone [19,21]. When compared to its beneficial effects on NASH, the effect of pioglitazone on liver fibrosis seems (rather) modest [15,17-23]. However, in a placebo-controlled RCT of 101 United States individuals with biopsy-proven NASH and T2DM or pre-diabetes, who were randomly assigned to

of

pioglitazone (45 mg/day) or placebo for 18 months, Cusi et al. reported that among those treated with

ro

pioglitazone, 51% had histological resolution of NASH [23]. Moreover, long-term pioglitazone treatment improved individual histologic scores of NASH, including the fibrosis score. Notably, all 18-month histologic

-p

improvements persisted over 36 months of pioglitazone treatment. The overall rate of adverse events did not differ between the two groups, although weight gain was greater with pioglitazone (~2.5 kg vs. placebo)

re

[23]. Interestingly, in a meta-analysis of eight RCTs (5 evaluating pioglitazone use and 3 evaluating

rn al P

rosiglitazone use) enrolling 516 adults with biopsy-confirmed NASH for a duration of 6 to 24 months, Musso et al. reported that pioglitazone improved advanced fibrosis in NASH, even in patients without diabetes [44]. Based on all the aforementioned data, the American and European guidelines recommended the use of pioglitazone in patients with biopsy-proven NASH, regardless of diabetes status [5,6,45]. However, it is important to highlight that pioglitazone is not yet approved by most national Medicines agencies outside the

Jo u

treatment for T2DM, and its off-label use for NAFLD/NASH treatment requires the patient’s consent. Concerns about weight gain, fluid retention and risk of bone fractures (especially in women) may restrict the wider clinical use of pioglitazone in all patients with NAFLD [43]. However, it is important to remember that pioglitazone may also exert some cardiovascular benefits to decrease risk of acute myocardial infarction and stroke in patients with T2DM or pre-diabetes [46,47]. Taking into consideration that it is now well accepted that patients with NASH are at higher risk of cardiovascular disease [1,5,6], and pioglitazone is an inexpensive, generic medication, this cardiovascular-protective agent should be considered in patients with NASH. Similarly, newer selective PPAR-gamma modulators such as CHRS 131 (a.k.a. INT 131), which retain pioglitazone like efficacy without many of the pioglitazone side effects [48], including weight gain, need to

11

Page 11 of 38

be studied carefully in the context of RCTs since they may prove to be a useful addition to our therapeutic armamentarium. DPP-4 inhibitors and GLP-1RAs are broadly prescribed as additional therapy in patients with T2DM [43]. It is important to underline that GLP-1 receptors have been documented in human hepatocytes and that the activation of such receptors may concur to decrease liver steatosis by improving insulin-signalling pathways [7,8]. In addition, GLP-1RAs induce significant weight loss (on average ~3-5 kg) [43]. For these reasons, GLP1RAs have also been investigated as a therapeutic option for NAFLD or NASH. Our systematic review supports

of

the capability of GLP-1RAs to reduce serum liver enzymes and improve hepatic steatosis, as detected by

ro

either imaging techniques or liver histology [24-29]. In particular, in the LEAN trial [25], patients (n=23) with biopsy-confirmed NASH who received liraglutide 1.8 mg/day for 48 weeks had a greater histological

-p

resolution of NASH with no worsening in hepatic fibrosis (39% vs. 9%) and significant improvements in

re

hepatic steatosis and hepatocyte ballooning compared with patients (n=22) receiving placebo. The authors suggested that the beneficial effects of liraglutide on the histological liver endpoints are possibly due to both

rn al P

its direct hepatic effect and to concomitant weight loss [25]. Importantly, liraglutide and other long-acting GLP-1RAs have been also shown to reduce risk of developing cardiovascular and renal outcomes in patients with T2DM [49]. For such reasons, if larger phase 3 RCTs will further confirm the promising findings of the LEAN trial, it is reasonable to hypothesize that liraglutide (and other long-acting GLP-1RAs) will become a

Jo u

suitable treatment option in NAFLD patients, especially in those who are obese or have T2DM. Currently, no robust data exist with liver histological endpoints as a primary outcome to comment on the effectiveness of DPP-4 inhibitors as a treatment for NAFLD. In addition, DPP-4 inhibitors also have a neutral effect on cardiovascular events in T2DM patients [43], thus making this class of anti-hyperglycaemic agents even less attractive for treatment of NAFLD. SGLT-2 inhibitors are a newer class of anti-hyperglycaemic agents that increase glucose reabsorption mainly by the kidney [43] v. SGLT-2 is expressed on renal epithelial cells that line the S1 segment of the proximal convoluted tubule, and SGLT-2 plays a key role in promoting glycosuria. In this way, this mechanism of regulation of blood glucose control is largely independent of insulin secretion [43]. Experimental animal

12

Page 12 of 38

studies support a beneficial effect of SGLT-2 inhibitors on liver steatosis, hepatocyte ballooning and, in some cases, also on liver fibrosis, possibly due to a combination of negative energy balance by increased glycosuria and substrate switching towards lipids as a source of energy expenditure [7,43]. Experimental animal studies also support a beneficial effect of SGLT-2 inhibitors on insulin resistance and lipotoxicity [7,43]. Our systematic review supports the possibility that SGLT-2 inhibitors may improve both serum liver enzyme levels and liver fat content, as assessed by imaging techniques. However, most of RCTs are small with a short period of treatment and, most importantly, to date, there are no head-to-head or placebo-controlled RCTs

of

examining the long-term effects of SGLT-2 inhibitors on histologic features of NAFLD [50]. By contrast, SGLT2-

ro

inhibitors have shown significant cardio-renal benefits in large RCTs of patients with T2DM [51]; and this may represent an attractive bonus for the use of these drugs in NAFLD patients [50]. Lastly, it is important to note

-p

there are multiple ongoing RCTs testing the effect of SGLT2-inhibitors in patients with NAFLD.

re

We believe that the major strength and the added value of our study are the use of systematic review processes to identify head-to-head or placebo-controlled RCTs (published until September 30, 2019) that

rn al P

meet pre-defined inclusion criteria. Limitations of this systematic review include restriction to RCTs, which may have limited generalizability to ‘real-world’ populations of patients with NAFLD, and also the highly variable mechanisms of actions of the anti-hyperglycaemic agents, which limits any conclusions about any potential association between anti-hyperglycaemic agents and improvement in NAFLD. Another limitation is

Jo u

the almost absence of head-to-head comparative RCTs (see Tables 1-4), making difficult a direct comparison between the different anti-hyperglycaemic agents. Finally, although these RCTs have included different ethnic groups, in none of these RCTs it has been examined the contribution of common genetic variants related to NAFLD (i.e., patatin-like phospholipase domain-containing protein 3 [PNPLA3] and transmembrane 6 superfamily member 2 [TM6SF2] polymorphisms) to the efficacy of anti-hyperglycaemic agents for treatment of NAFLD or NASH. Further studies are needed to better elucidate this issue. In addition, since sex differences do exist in the prevalence, risk factors and clinical outcomes of NAFLD [52], future studies should also be specifically designed to explore sex differences in the response to treatment for NAFLD or NASH with anti-hyperglycaemic agents.

13

Page 13 of 38

In conclusion, our systematic review is the most comprehensive and updated assessment of published headto-head, or placebo-controlled RCTs, of individuals with NAFLD that used an EMA-approved antihyperglycaemic drug to specifically treat NAFLD or NASH. Although it has been shown that many of the antihyperglycaemic agents improve serum liver enzyme levels, convincing data on their beneficial effects on histologic features of NAFLD are very limited. Our systematic review supports the conclusion that most of the available evidence of efficacy in patients with NASH relates to the use of pioglitazone. Not only might pioglitazone also improve the natural history of liver disease by reducing its progression to cirrhosis in some

of

patients with biopsy-confirmed NASH, but there is proven evidence that long-term pioglitazone treatment

ro

may also decrease risk of incident cardiovascular events, such as myocardial infarction and ischemic stroke. That said, long-term safety concerns from RCTs in patients with T2DM (and which have not been shown in

-p

patients with NASH), are undoubtedly limiting pioglitazone usage in clinical practice. We suggest that further well-designed RCTs with longer follow-up and adequate liver histology endpoints are needed to better

re

characterize the efficacy and harms of this generic and inexpensive medication in patients with NAFLD. Larger

rn al P

phase 3 RCTs with adequate liver histology endpoints are also needed to confirm the long-term beneficial effects of GLP1 RAs and SGLT-2 inhibitors, considering their promising effects on NAFLD using magnetic resonance or other imaging techniques. Finally, we believe that tailoring pharmacotherapy to the dominant pathogenic pathway in a given patient, along with the use of combination therapies, is likely to represent the

Jo u

future direction in the treatment of patients with NASH (irrespective of the presence of diabetes).

14

Page 14 of 38

ACKNOWLEDGEMENTS Disclosure Statement: nothing to declare, except for Mantzoros CS, who is a shareholder and has received honoraria and grant support through his Institution (BIDMC) from Novo Nordisk and Coherus Inc. Funding: GT is supported in part by grants from the University School of Medicine of Verona, Verona, Italy. CDB is supported in part by grants from the Southampton National Institute for Health Research Biomedical Research Centre. Authors’ Contributions: AM and GT conceived and designed the study. AM, CDB and GT researched data and

of

reviewed/edited the manuscript. ES and CSM contributed to discussion and reviewed/edited the manuscript.

ro

AM and GT analyzed the data. GT and CDB wrote the manuscript. GT is the guarantor of this work and, as such, had full access to all the data of the study and take responsibility for the integrity and accuracy of data.

-p

All authors approved the final version of the manuscript.

re

Appendix supplementary material

Supplementary materials (Fig. S1 and Table S1) associated with this article can be found at

FIGURE LEGEND

rn al P

http://www.scincedirect.com at doi . . .

Jo u

Supplementary Figure S1. Flow-chart of the literature research and study selection.

15

Page 15 of 38

REFERENCES

Jo u

rn al P

re

-p

ro

of

1. Lonardo A, Bellentani S, Argo CK, Ballestri S, Byrne CD, Caldwell SH, et al. Epidemiological modifiers of non-alcoholic fatty liver disease: focus on high-risk groups. Dig Liver Dis 2015;47:997-1006. 2. Younossi ZM, Golabi P, de Avila L, Paik JM, Srishord M, Fukui N, et al. The global epidemiology of NAFLD and NASH in patients with type 2 diabetes: A systematic review and meta-analysis. J Hepatol 2019;71:793-801. 3. Ballestri S, Zona S, Targher G, Romagnoli D, Baldelli E, Nascimbeni F, et al. Nonalcoholic fatty liver disease is associated with an almost twofold increased risk of incident type 2 diabetes and metabolic syndrome. Evidence from a systematic review and meta-analysis. J Gastroenterol Hepatol 2016;31:936-44. 4. Mantovani A, Byrne CD, Bonora E, Targher G. Nonalcoholic fatty liver disease and risk of incident type 2 diabetes mellitus: a meta-analysis. Diabetes Care 2018;41:372-82. 5. Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, et al. The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases. Hepatology 2018;67:328-57. 6. European Association for the Study of the Liver (EASL); European Association for the Study of Diabetes (EASD); European Association for the Study of Obesity (EASO). EASL-EASD-EASO Clinical Practice Guidelines for the management of non-alcoholic fatty liver disease. J Hepatol 2016;64:1388-402. 7. Raschi E, Mazzotti A, Poluzzi E, De Ponti F, Marchesini G. Pharmacotherapy of type 2 diabetes in patients with chronic liver disease: focus on nonalcoholic fatty liver disease. Expert Opin Pharmacother 2018;19:1903-14. 8. Cholankeril R, Patel V, Perumpail BJ, Yoo ER, Iqbal U, Sallam S, et al. Anti-diabetic medications for the pharmacologic management of NAFLD. Diseases 2018;6. pii:E93. 9. McDonagh MS, Jonas DE, Gartlehner G, Little A, Peterson K, Carson S, et al. Methods for the drug effectiveness review project. BMC Med Res Methodol. 2012;12:140. 10. Harris RP, Helfand M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, et al. Current methods of the US Preventive Services Task Force: a review of the process. Am J Prev Med 2001,20:2135. 11. Bugianesi E, Gentilcore E, Manini R, Natale S, Vanni E, Villanova N, et al. A randomized controlled trial of metformin versus vitamin E or prescriptive diet in nonalcoholic fatty liver disease. Am J Gastroenterol 2005;100:1082-90. 12. Haukeland JW, Konopski Z, Eggesbø HB, von Volkmann HL, Raschpichler G, Bjøro K, et al. Metformin in patients with non-alcoholic fatty liver disease: a randomized, controlled trial. Scand J Gastroenterol 2009;44:853-60. 13. Omer Z, Cetinkalp S, Akyildiz M, Yilmaz F, Batur Y, Yilmaz C, et al. Efficacy of insulin-sensitizing agents in nonalcoholic fatty liver disease. Eur J Gastroenterol Hepatol 2010;22:18-23. 14. Lavine JE, Schwimmer JB, Van Natta ML, Molleston JP, Murray KF, Rosenthal P, et al. Effect of vitamin E or metformin for treatment of nonalcoholic fatty liver disease in children and adolescents: the TONIC randomized controlled trial. JAMA 2011;305:1659-68. 15. Razavizade M, Jamali R, Arj A, Matini SM, Moraveji A, Taherkhani E. The effect of pioglitazone and metformin on liver function tests, insulin resistance, and liver fat content in nonalcoholic fatty liver disease: a randomized double blinded clinical trial. Hepat Mon 2013;13:e9270. 16. Rana H, Yadav SS, Reddy HD, Singhal S, Singh DK, Usman K. Comparative effect of insulin sensitizers and statin on metabolic profile and ultrasonographical score in non-alcoholic fatty liver disease. J Clin Diagn Res 2016;10:OC19-23. 16

Page 16 of 38

Jo u

rn al P

re

-p

ro

of

17. Belfort R, Harrison SA, Brown K, Darland C, Finch J, Hardies J, et al. A placebo-controlled trial of pioglitazone in subjects with nonalcoholic steatohepatitis. N Engl J Med 2006;355:2297307. 18. Aithal GP, Thomas JA, Kaye PV, Lawson A, Ryder SD, Spendlove I, et al. Randomized, placebocontrolled trial of pioglitazone in nondiabetic subjects with nonalcoholic steatohepatitis. Gastroenterology 2008;135:1176-84. 19. Ratziu V, Giral P, Jacqueminet S, Charlotte F, Hartemann-Heurtier A, Serfaty L, et al. Rosiglitazone for nonalcoholic steatohepatitis: one-year results of the randomized placebocontrolled fatty liver improvement with rosiglitazone therapy (FLIRT) trial. Gastroenterology 2008;135:100-10. 20. Sanyal AJ, Chalasani N, Kowdley KV, McCullough A, Diehl AM, Bass NM, et al. Pioglitazone, vitamin E, or placebo for nonalcoholic steatohepatitis. N Engl J Med 2010;362:1675-85. 21. Torres DM, Jones FJ, Shaw JC, Williams CD, Ward JA, Harrison SA. Rosiglitazone versus rosiglitazone and metformin versus rosiglitazone and losartan in the treatment of nonalcoholic steatohepatitis in humans: a 12-month randomized, prospective, open- label trial. Hepatology 2011;54:1631-9. 22. Sharma BC, Kumar A, Garg V, Reddy RS, Sakhuja P, Sarin SK. A randomized controlled trial comparing efficacy of pentoxifylline and pioglitazone on metabolic factors and liver histology in patients with non-alcoholic steatohepatitis. J Clin Exp Hepatol 2012;2:333-7. 23. Cusi K, Orsak B, Bril F, Lomonaco R, Hecht J, Ortiz-Lopez C, et al. Long-term pioglitazone treatment for patients with nonalcoholic steatohepatitis and prediabetes or type 2 diabetes mellitus: a randomized trial. Ann Intern Med 2016;165:305-15. 24. Shao N, Kuang HY, Hao M, Gao XY, Lin WJ, Zou W. Benefits of exenatide on obesity and nonalcoholic fatty liver disease with elevated liver enzymes in patients with type 2 diabetes. Diabetes Metab Res Rev 2014;30:521-9. 25. Armstrong MJ, Gaunt P, Aithal GP, Barton D, Hull D, Parker R, et al. Liraglutide safety and efficacy in patients with non-alcoholic steatohepatitis (LEAN): a multicentre, double-blind, randomised, placebo-controlled phase 2 study. Lancet 2016;387:679-90. 26. Dutour A, Abdesselam I, Ancel P, Kober F, Mrad G, Darmon P, et al .Exenatide decreases liver fat content and epicardial adipose tissue in patients with obesity and type 2 diabetes: a prospective randomized clinical trial using magnetic resonance imaging and spectroscopy. Diabetes Obes Metab 2016;18:882-91. 27. Feng W, Gao C, Bi Y, Wu M, Li P, Shen S, et al. Randomized trial comparing the effects of gliclazide, liraglutide, and metformin on diabetes with non-alcoholic fatty liver disease. J Diabetes 2017;9:800-9. 28. Frøssing S, Nylander M, Chabanova E, Frystyk J, Holst JJ, Kistorp C, et al. Effect of liraglutide on ectopic fat in polycystic ovary syndrome: a randomized clinical trial. Diabetes Obes Metab 2018;20:215-8. 29. Yan J, Yao B, Kuang H, Yang X, Huang Q, Hong T, et al. Liraglutide, sitagliptin, and insulin glargine added to metformin: the effect on body weight and intrahepatic lipid in patients with type 2 diabetes mellitus and nonalcoholic fatty liver disease. Hepatology 2019;69:241426. 30. Macauley M, Hollingsworth KG, Smith FE, Thelwall PE, Al-Mrabeh A, Schweizer A, et al. Effect of vildagliptin on hepatic steatosis. J Clin Endocrinol Metab 2015;100:1578-85. 31. Cui J, Philo L, Nguyen P, Hofflich H, Hernandez C, Bettencourt R, et al. Sitagliptin vs. placebo for non-alcoholic fatty liver disease: a randomized controlled trial. J Hepatol 2016;65:36976.

17

Page 17 of 38

Jo u

rn al P

re

-p

ro

of

32. Deng XL, Ma R, Zhu HX, Zhu J. Short article: a randomized-controlled study of sitagliptin for treating diabetes mellitus complicated by nonalcoholic fatty liver disease. Eur J Gastroenterol Hepatol 2017;29:297-301. 33. Bolinder J, Ljunggren Ö, Kullberg J, Johansson L, Wilding J, Langkilde AM, et al. Effects of dapagliflozin on body weight, total fat mass, and regional adipose tissue distribution in patients with type 2 diabetes mellitus with inadequate glycemic control on metformin. J Clin Endocrinol Metab 2012;97:1020-31. 34. Ito D, Shimizu S, Inoue K, Saito D, Yanagisawa M, Inukai K, et al. Comparison of ipragliflozin and pioglitazone effects on nonalcoholic fatty liver disease in patients with type 2 diabetes: a randomized, 24-week, open-label, active-controlled trial. Diabetes Care 2017;40:1364-72. 35. Cusi K, Bril F, Barb D, Polidori D, Sha S, Ghosh A, et al. Effect of canagliflozin treatment on hepatic triglyceride content and glucose metabolism in patients with type 2 diabetes. Diabetes Obes Metab 2018 Nov 16. doi: 10.1111/dom.13584 [Epub ahead of print]. 36. Kuchay MS, Krishan S, Mishra SK, Farooqui KJ, Singh MK, Wasir JS, et al. Effect of empagliflozin on liver fat in patients with type 2 diabetes and nonalcoholic fatty liver disease: a randomized controlled trial (E-LIFT trial). Diabetes Care 2018;41:1801-8. 37. Eriksson JW, Lundkvist P, Jansson PA, Johansson L, Kvarnström M, Moris L, et al. Effects of dapagliflozin and n-3 carboxylic acids on non-alcoholic fatty liver disease in people with type 2 diabetes: a double-blind randomised placebo-controlled study. Diabetologia 2018;61:1923-34. 38. Shimizu M, Suzuki K, Kato K, Jojima T, LIijima T, Murohisa T, et al. Evaluation of the effects of dapagliflozin, a sodium-glucose cotransporter-2 inhibitor, on hepatic steatosis and fibrosis using transient elastography in patients with type 2 diabetes and non-alcoholic fatty liver disease. Diabetes Obes Metab 2019;21:285-92. 39. Kahl S, Gancheva S, Straßburger K, Herder C, Machann J, Katsuyama H, et al. Empagliflozin effectively lowers liver fat content in well-controlled type 2 diabetes: a randomized, doubleblind, phase 4, placebo-controlled trial. Diabetes Care 2019 Sep 20. pii: dc190641. doi: 10.2337/dc19-0641. [Epub ahead of print]. 40. Cheung A, Neuschwander-Tetri BA, Kleiner DE, Schabel E, Rinella M, Harrison S, et al. Defining improvement in nonalcoholic steatohepatitis for treatment trial endpoints: recommendations from the liver forum. Hepatology 2019;70:1841-55. 41. Ratziu V. A critical review of endpoints for non-cirrhotic NASH therapeutic trials. J Hepatol 2018;68:353-61. 42. Adams LA. End-points for drug treatment in NASH. Hepatol Int 2019;13:252-8. 43. Pharmacologic approaches to glycemic treatment: standards of medical care in diabetes 2018. Diabetes Care 2018;41:S73-S85. 44. Musso G, Cassader M, Paschetta E, Gambino R. Thiazolidinediones and advanced liver fibrosis in nonalcoholic steatohepatitis: a meta-analysis. JAMA Intern Med 2017;177:633-40. 45. Glen J, Floros L, Day C, Pryke R. Non-alcoholic fatty liver disease (NAFLD): summary of NICE guidance. BMJ 2016;354:i4428. 46. Dormandy JA, Charbonnel B, Eckland DJ, Erdmann E, Massi-Benedetti M, Moules IK, et al. Secondary prevention of macrovascular events in patients with type 2 diabetes in the PROactive Study (PROspective pioglitAzone Clinical Trial In macroVascular Events): a randomised controlled trial. Lancet 2005;366:1279-89. 47. Kernan WN, Viscoli CM, Furie KL, Young LH, Inzucchi SE, Gorman M, Guarino PD, et al. Pioglitazone after ischemic stroke or transient ischemic attack. N Engl J Med 2016;374:132131.

18

Page 18 of 38

of

48. Athyros VG, Polyzos SA, Kountouras J, Katsiki N, Anagnostis P, Doumas M, et al. Nonalcoholic fatty liver disease treatment in patients with type 2 diabetes mellitus; new kids on the block. Curr Vasc Pharmacol 2019 Apr 5. doi: 10.2174/1570161117666190405164313 [Epub ahead of print]. 49. Kristensen SL, Rørth R, Jhund PS, Docherty KF, Sattar N, Preiss D, et al. Cardiovascular, mortality, and kidney outcomes with GLP-1 receptor agonists in patients with type 2 diabetes: a systematic review and meta-analysis of cardiovascular outcome trials. Lancet Diabetes Endocrinol 2019;7:776-85. 50. Scheen AJ. Beneficial effects of SGLT2 inhibitors on fatty liver in type 2 diabetes: a common comorbidity associated with severe complications. Diabetes Metab 2019;45:213-23. 51. Kluger AY, Tecson KM, Lee AY, Lerma EV, Rangaswami J, Lepor NE, et al. Class effects of SGLT2 inhibitors on cardiorenal outcomes. Cardiovasc Diabetol 2019;18:99. 52. Lonardo A, Nascimbeni F, Ballestri S, Fairweather D, Win S, Than TA, et al. Sex differences in nonalcoholic fatty liver disease: state of the art and identification of research gaps. Hepatology 2019;70:1457-69.

Jo u

rn al P

re

-p

ro

Figure Caption

19

Page 19 of 38

Table I. Placebo-controlled or active-controlled RCTs of metformin for treatment of NAFLD (ordered by publication year).

Non-diabetic adults with biopsy-confirmed NAFLD Age: 43 y

Interventions (group sizes), Duration A: Metformin 2000 mg/d (n = 55) B: Vitamin E 400 IU/d (n = 28)

Sex: 85% male C: Prescriptive diet (n = 27)

BMI, kg/m2: 28.5 Diabetes: 8.1% diagnosed

newly

Sex: 73% male

B: Placebo (n = 24)

Ethnicity: 86.4% white

Duration: 6 months

re

rn al P

Adults with biopsyconfirmed NAFLD

BMI, kg/m2: 30.8 Diabetes: 27%

Jo u

Mean AST and ALT: NR

Omer, 2010 Turkey (Fair)

(13)

Serum aminotransferase levels improved in all groups, in association with weight loss. The effect in the metformin arm was larger (P < 0.0001), and serum ALT levels normalized in 56% of cases

No patients on metformin or vitamin E stopped treatments because of AEs

A control biopsy in 17 metformin-treated cases (14 non-responders) showed a significant decrease in liver fat (P < 0.001), necroinflammation, and fibrosis (P = 0.012 for both) Percentage with improvement (P -value change from baseline); between-groups p-value:

Serious AEs: NR Withdrawal due to AEs: 2/24 (8.3%) vs. 0/24 (0%)

Steatosis: 25% (P = 0.10) vs. 38% (P = 0.03); P = 0.052 Fibrosis: 5% (P = 0.99) vs. 17% (P = 0.17); P = 0.36 NAFLD activity score: 20% (P = 0.23) vs. 50% (P = 0.12); P = 0.06 Changes from baseline (Pvalue); between-groups Pvalue: Weight: -4.3 (P < 0.001) vs. 0.3 kg (P = 0.45); P < 0.001

Adults with type 2 diabetes or impaired glucose tolerance and biopsy-confirmed NAFLD

A: Metformin 1700 mg/d plus rosiglitazone 4 mg/d (n = 22)

Age: 48.9 y

B: Metformin 1700 mg/d (n = 22)

Sex: 54.7% male BMI, kg/m2: 30.6

Adverse effects

-p

Age: 47.4 y

A second liver biopsy was programmed at the end of the study, but was considered optional A: Metformin 2500 mg/d (3000 mg if weight >90 kg) (n = 24)

Mean AST 45 U/L, ALT 90 U/L

Haukeland, 2009 (12) Norway (Good)

Duration: 12 months

Efficacy/effectiveness outcomes A vs. B (vs. C)

of

Population, Demographics

ro

Investigator, Year, Country, Trial name (Quality rating) Bugianesi, 2005 (11) multicentre Italy (Fair)

C: Rosiglitazone 4 mg/d (n = 20)

BMI: -1.3 (P < 0.001) vs. 0.1 kg/m2 (P = 0.59); P < 0.001 Changes from baseline (pvalue):

NR

NAFLD score: -3.9 (P = 0.026) vs. 0.7 (P = 0.73) vs. -2.6 (P = 0.012) AST: -15.4 (P = 0.01) vs. 13.0 (P = NS) vs. -13.2 UI/L (P = 0.005)

20

Page 20 of 38

Diabetes: NR

Duration: 12 months

Mean AST 50 UI/L, ALT 67 UI/L

Lavine, 2011 (14), multicenter United States, TONIC trial (Good)

Children/adolescents with biopsy-proven NAFLD

A. Metformin 1000 mg/d (n = 57)

Age: 13.1 y

B. Vitamin E 800 IU/d (n = 58)

Sex: 81% male

C. Placebo (n = 58)

Ethnicity: White 74%

Duration: 96 weeks

ro

NASH: 42%

re rn al P Jo u

Adults with NAFLD assessed by ultrasonography and predictive formula

A: Metformin 1000 mg/d (n = 40) B: Pioglitazone 30 mg/d (n = 40)

Age: 35.3 y Duration: 4 months Sex: 85% male BMI, kg/m2: 27.7 Diabetes: 7.5%

Change in hepatocellular ballooning scores: +0.1 with placebo (95% CI, -0.2 to 0.3) vs. -0.5 with vitamin E (95% CI, -0.8 to -0.3; P = 0.006) vs. -0.3 with metformin (95% CI, -0.6 to -0.0; P = 0.04); and in NAFLD activity score, -0.7 with placebo (95% CI, -1.3 to -0.2) vs. -1.8 with vitamin E (95% CI, -2.4 to 1.2; P = 0.02) and -1.1 with metformin (95% CI, -1.7 to -0.5; P = 0.25)

-p

Mean AST 71 UI/L, ALT 123 UI/L

2013 Iran

NR

Change in ALT level from baseline at week 96: -41.7 vs -48.3 vs. -35.2 IU/L, p=NS

Diabetes: 0%

Razavizade, (15) (Fair)

BMI: -1.3 (P = 0.006) vs. 3.2 (P = 0.002) vs. -0.3 kg/m2 (P = 0.29) Neither vitamin E nor metformin was superior to placebo in attaining the primary outcome of sustained reduction in ALT level or the secondary endpoint (improvements in histological features of NAFLD and resolution of NASH)

of

BMI, kg/m2: 34

ALT: -22.7 (P = 0.017) vs. 16.7 (P = 0.62) vs. -36.2 UI/L (P < 0.0001)

Resolution of NASH: 41% vs. 58% vs. 28%, P = NS There was an overall mean increase in weight and BMI, but there were no significant differences Changes from baseline (Pvalue), between-groups pvalue:

Serious AEs: NR Withdrawal due to AEs: None

Liver fat fraction (on ultrasound): -2.53 (P < 0.01) vs. -3.23 (P < 0.01), P = 0.48 AST: -10.8 (P < 0.01) vs. 13.8 UI/L (P < 0.01), P = 0.56

Mean AST and ALT: NR

21

Page 21 of 38

ALT: -21.8 (P < 0.01) vs. 37.5 UI/L (P < 0.01), P = 0.07

Rana, 2016 India (Fair)

(16)

Adults with ultrasounddetected NAFLD without history of use of insulin sensitizers or hypolipidemic drugs

A. Metformin (n = 31)

Age: NR

C. Pioglitazone (n = 33)

B. Rosuvastatin (n = 34)

Sex: NR

Weight: -2.7 (P < 0.01) vs. 1.2 kg (P = 0.04), P = 0.05 Change in ultrasound score (fatty liver) at 24 weeks: our analysis

NR

A vs. B: 0.07 vs. -1.27 (P < 0.001) A vs. C: 0.07 vs. -0.70 (P < 0.001)

Duration: 24 weeks Weight change at weeks: our analysis

24

of

NB: daily dosage of three drugs was not reported

A vs. B: -4.8 vs. -4.3 kg (P = 0.13) A vs. C: -4.8 vs. 0.03 kg (P < 0.001)

ro

Mean AST 55 IU/mL, ALT 64 UI/mL (AST and ALT were different at baseline between treatment groups) BMI, kg/m2: 27.9

-p

AST change at 24 weeks: our analysis

rn al P

re

Diabetes: NR

A vs. B: -14.1 vs. 8.4 UI/L (P < 0.001) A vs. C: -14.1 vs. 23.7 UI/L (P = 0.04) ALT change at 24 weeks: our analysis A vs. B: -15.6 vs. 8.1 UI/L (P < 0.001) A vs. C: -15.6 vs. 24.7 UI/L (P = 0.13)

Jo u

Abbreviations: AEs, adverse effects; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; NR, not reported; NS, not significant.

22

Page 22 of 38

Table II. Placebo-controlled or active-controlled RCTs of glitazones for treatment of NAFLD (ordered by publication year). Investigator, Year, Country, Trial name (Quality rating) Belfort, 2006 (17) US (Good)

Population, Demographics

Interventions (group sizes), Duration

Efficacy/effectiveness outcomes A vs. B (vs. C)

Adverse effects

Adults with type 2 diabetes or impaired glucose tolerance and biopsy-confirmed NASH

A: Pioglitazone 30 mg/d for 2 months, then 45 mg/d (n = 29)

Percent with liver fibrosis improvement: 46% vs. 33%, P = 0.08

Serious AEs: NR

B: Placebo (n = 25)

Age: 51 y

Duration: 6 months

Changes from baseline (pvalue), between-groups pvalue:

Sex: 45% male

Withdrawal due to AEs: 1/29 (3.5%) vs. 1/25 (4.0%)

AST: -19 (P < 0.001) vs. -9 UI/L (P = 0.08), P = 0.04

Ethnicity: NR

of

ALT: -39 (P < 0.001) vs. -21 UI/L (P = 0.03), P < 0.001

HbA1: 6.2%

Weight: 2.5 (P < 0.001) vs. 0.5 kg (P = 0.53), P = 0.003

ro

BMI, kg/m2: 33.2 Diabetes: NR

-p

A: Pioglitazone 30 mg/d (n = 37)

Number (%) with improvement (P- value), between-groups P-value:

re

Aithal, 2008 (18) UK (Good)

Mean AST 44 U/L, ALT 64 U/L Non-diabetic adults with biopsy- confirmed NASH

BMI: 1.1 (P < 0.001) vs. -0.2 kg/m2 (P = 0.62), P = 0.005

B: Placebo (n = 37)

rn al P

Age: 53 y

Duration: 12 months

Sex: 61% male

Ethnicity: white

Jo u

BMI, kg/m2: 30.3

Ratziu, 2008 (19) France FLIRT (Good)

Adults with biopsyconfirmed NASH

Liver fibrosis: 9/31 (29%) (P = 0.006) vs. 6/30 (20%) (P = 0.81), P = 0.05

Serious AEs: NR Withdrawal due to AEs: 3/37 (8.1%) vs. 4/37 (10.8%)

Steatosis: 15/31 (48%) (P = 0.001) vs. 11/30 (37%) (P = 0.03), P = 0.19 Changes from baseline (Pvalue), between-groups pvalue: Weight: 2.6 (P = 0.005) vs. 3.5 kg (P = 0.69), P = 0.02

A: Rosiglitazone 8 mg/d (4 mg/d for 1st month) (n = 32)

Age: 53.6 y B: Placebo (n = 31)

ALT: -37.7 (p=0.02) vs. -6.9 UI/L (p=0.41), p=0.01 Changes from baseline, between- groups p-value: NAFLD activity score: -1 vs. 0, P = 0.60

Serious AEs: NR Withdrawal due to AEs: 1/32 (3.1%) vs. 0/31

Sex: 59% male Duration: 12 months Ethnicity: NR BMI, kg/m2: 31

Steatosis, % reduction: 20% vs. - 5%, P = 0.02

Dose reduction due to AEs: 5/32 (15.6%) vs. 1/31 (3.2%)

Fibrosis: 0.03 vs. -0.18, P = 0.43

23

Page 23 of 38

Diabetes: 32%

ALT, number (%) achieving normalization: 12/32 (38%) vs. 2/31 (7%), P = 0.005 ALT, mean % change from baseline: -28% vs. -2%; mean reduction, -44% vs. 0%, P < 0.001

A: Pioglitazone 30 mg/d (n = 80)

Age: 46.3 y

B: Vitamin E 800 IU/d (n = 84)

Sex: 40% male

C: Placebo (n = 83)

Ethnicity, % white: 88

Duration: 96 weeks

NASH improvement, n (%): 27/80 (34%) (P = 0.04) vs. 36/84 (43%) (P = 0.001) vs. 16/83 (19%)

Serious AEs: NR Withdrawal due to AEs: None

of

Non-diabetic adults with biopsy- confirmed NASH

ro

Sanyal, 2010 (20) US PIVENS (Good)

AST, mean % change from baseline: -8% vs. 9%; mean reduction, -62% vs. +15%, P < 0.001 Changes from baseline (pvalue vs. placebo):

NAFLD activity score: -1.9 (P < 0.001) vs. -1.9 (P < 0.001) vs. -0.5

-p

BMI, kg/m2: 34

Steatosis: -0.8 (P < 0.001) vs. -0.7 (P < 0.001) vs. -0.1

re

Mean AST 56 U/L, ALT 83 U/L

AST: -20.4 (P < 0.001) vs. 21.3 (P < 0.001) vs. -3.8 UI/L ALT: -40.8 (P < 0.001) vs. 37.0 (P = 0.001) vs. -20.1 UI/L

Adults with biopsyconfirmed NASH

A: Rosiglitazone mg/d (n = 50)

Age: 49.4 y

B: Rosiglitazone 8 mg/d + metformin 1000 mg/d (n = 50)

Jo u

Torres, 2011 (21) US (Good)

rn al P

Fibrosis: -0.4 (P = 0.10) vs. 0.3 (P = 0.19) vs. -0.1

Sex: 44% male Ethnicity, %: Caucasian: 65 BMI, kg/m2: 33.2 Diabetes: 17% HbA1c: 5.9%

8

C: Rosiglitazone 8 mg/d + losartan 50 mg/d (n = 50) Duration: 48 weeks

Weight: 4.7 (P < 0.001) vs. 0.4 (P = 0.65) vs. 0.7 kg Subjects with final biopsy: 26 vs. 28 vs. 35 Changes from baseline, between- groups P-value:

Serious AEs: NR Withdrawal due to AEs: NR by group

Resolution of definite NASH, n (%): 12/26 (46%) vs. 10/28 (36%) vs. 10/35 (29%), P = NR NAFLD activity score: -1.77 vs. -1.32 vs. -1.37, P = 0.67 Steatosis: -0.85 vs. -0.82 vs. -0.74, P = 0.91

Mean AST 56 U/L, ALT 86 U/L

24

Page 24 of 38

Fibrosis: -0.70 vs. -0.59 vs. 0.32, P = 0.30 AST: -39.6 vs. -35.0 vs. 48.7 UI/L, P = NS (exact pvalue NR) ALT: -17.4 vs. -19.9 vs. 21.7 UI/L, P = NS (exact pvalue NR)

Adults with biopsyconfirmed NASH

A: Pentoxifylline 1200 mg/d (n = 29)

Age: 38.9 y

B: Pioglitazone 30 mg/d (n = 30)

Sex: 54% male Duration: 24 weeks

-p

Diabetes: NR

re

2013 Iran

Fibrosis: 0.08 (P = 0.70) vs. -0.46 (P = 0.19), P = 0.26

A: Metformin 1000 mg/d (n = 40)

rn al P

Razavizade, (15) (Fair)

Age: 35.3 y

B: Pioglitazone 30 mg/d (n = 40)

Sex: 85% male

Duration: 4 months

BMI, kg/m2: 27.7 Diabetes: 7.5%

Jo u

Mean AST 50 U/L, ALT 91 U/L

Cusi, 2016 (23) United States (Good)

Patients with type 2 diabetes or prediabetes and biopsy-confirmed NASH Age: 50.5 y Sex: 70.3 % male Ethnicity: Hispanic

67.3%

Brunt’s score: -0.34 (P = 0.10) vs. -1.2 (P = 0.005), P = 0.04

Steatosis: -0.83 (P = 0.02) vs. -1.18 (P = 0.005), P = 0.60

BMI, kg/m2: 24.9

Mean AST 65 U/L, ALT 96 U/L Adults with NAFLD assessed by ultrasonography

Withdrawal due to AEs: None

ro

Ethnicity: NR

Serious AEs: NR

of

Sharma, 2012 (22) India (Fair)

Weight: 0.9 vs. -1.2 vs. 3.7 kg, P = 0.051 Changes from baseline (Pvalue), between-groups Pvalue:

A. Pioglitazone 45 mg per day (n = 50) B. Placebo, (n = 51) All patients were prescribed a hypocaloric diet. Both groups followed with an open-label phase with pioglitazone for 18 months

Changes from baseline (Pvalue), between-groups Pvalue:

Serious AEs: NR Withdrawal due to AEs: None

Liver fat fraction: -2.53 (P < 0.01) vs. - 3.23 (P < 0.01), P = 0.48 AST: -10.8 (P < 0.01) vs. 13.7 UI/L (P < 0.01), P = 0.56 ALT: -21.7 (P < 0.01) vs. 37.5 UI/L (P < 0.01), P = 0.07 Weight: -2.7 (P < 0.01) vs. 1.2 kg (P = 0.04), P = 0.05 Greater than 2-point reduction of NAS without worsening fibrosis: 58% vs. 17%, P < 0.001

NR

NASH resolution: 51% vs. 19%, P < 0.001 Fibrosis; greater than 1point improvement: 39% vs. 25%, P = 0.13 Fibrosis mean change in score improved with

25

Page 25 of 38

BMI, kg/m2: 34.4 Diabetes: 51%

Duration: 18 months (36 months for openlabel phase)

pioglitazone: 0 vs. - 0.5, P < 0.05 Weight: pioglitazone group gained 2.5 kg, P = 0.039

HbA1c in those with diabetes (n = 51): 6.9%, or those with prediabetes: 5.7% (n = 50) Mean ALT 59 UI/L

Jo u

rn al P

re

-p

ro

of

Abbreviations: AE, adverse effects; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; NR, not reported.

26

Page 26 of 38

Page 27 of 38

27

of

ro

-p

re

rn al P

Jo u

Table III. Placebo-controlled or active-controlled RCTs of incretin-based therapies (GLP-1 receptor agonists or DPP-4 inhibitors) for treatment of NAFLD (ordered by publication year). Investigator, Year, Country, Trial name (Quality rating)

Population, Demographics

Interventions (group sizes), Duration

Efficacy/effectiveness outcomes A vs. B (vs. C vs. D)

Adverse effects

A. Exenatide glargine (n = 30)

+

Reversal rate of NAFLD based on liver ultrasound:

NR

A vs. B: 93% vs. 67%, P < 0.01

Age: 43 y

B. Intensive insulin: Insulin aspart + insulin glargine (n = 30)

Sex: 48% male

Duration: 12 weeks

GLP-1 receptor agonists 2014

(24)

Obese patients with type 2 diabetes and ultrasound-defined NAFLD (and raised serum liver enzymes)

Differences in weight change post minus pretreatment:

of

Shao, China (Fair)

BMI, kg/m2: 30

A vs. B: -7.8 vs. 3.3 kg, P < 0.001

ro

HbA1c: 7.6%

No difference between groups in change in HbA1c (-1.4 vs. -1.3%)

2016 United LEAN

Patients had biopsyproven noncirrhotic NASH

A. Liraglutide 1.8 mg (n = 26)

Resolution of NASH: 39% vs. 9% (relative risk 4.3, 95% CI 1.0 to 17.7)

re

Armstrong, (25) Kingdom, (Good)

-p

Mean ALT 166 IU/L, AST 123 IU/L

B. Placebo (n = 26)

rn al P

Age: 51 y

Duration: 48 weeks

Change in NAS score: -1.3 vs. -0.8, P = 0.24

Sex: 60% male

Ethnicity: White: 88% ALT: 72 IU/mL

liver

Jo u

F3-F4 on histology: 52%

BMI, kg/m2: 36 Diabetes: 33%

Patients with type 2 diabetes treated with metformin and/or sulphonylureas and/or DPP4 inhibitors (95% had NAFLD on MR spectroscopy) Age: 52 y

SAE: 8% vs. 8% GI-disorders: 81% vs. 65% (P = 0.27)

Change in fibrosis stage: 0.2 vs. 0.2, P = 0.11 Patients with improvement in fibrosis: 26% vs. 14%, P = 0.46 Patients with worsening fibrosis: 9% vs. 36%, P = 0.04 Change in ALT: -26.6 vs. 10.2 UI/L, P = 0.16

Mean ALT 71 IU/L, AST 51 IU/L

Dutour, 2016 (26), France (Fair)

WAE: 8% vs. 4% (P = 0.56)

Change in AST: -27 vs.+9 IU/L; P = 0.02 A: Placebo (n = 22) B: Exenatide 5-10 mcg bid (n = 22) Duration: 26 weeks

Exenatide and reference treatment led to a similar improvement in HbA1c (−0.7 ± 0.3% vs. −0.7 ± 0.4%; P = 0.29)

19 patients concluded the trial both in the placebo and in the treatment arm

Significant weight loss was observed only in the exenatide group (−5.5 ± 1.2 kg vs. −0.2 ± 0.8 kg; P =

28

Page 28 of 38

Sex: 48% male

0.001 for difference between groups)

BMI, kg/m2: 36.1

Feng, 2017 China (Fair)

(27)

Patients with type 2 diabetes and NAFLD assessed by ultrasonography

A. Liraglutide up to 1.8 mg/d (n = 31) B. Metformin up to 2000 mg/d (n = 31)

Age: 47 y Sex: 75% male

C. Gliclazide up to 120 mg/d (n = 31)

BMI, kg/m2: 27.6

Duration: 24 weeks

HbA1c: 9.1%

29 patients in each study arm completed the 24week trial

of

Mean ALT 29 IU/L, AST 22 IU/L

ro

HbA1c 7.5%

Exenatide induced a significant reduction in hepatic fat content, compared with the reference treatment (hepatic fat content: −23.8 ± 9.5% vs. +12.5 ± 9.6%, P = 0.007) Hepatic fat content (estimated by ultrasound) decreased significantly in all treatment groups, from 36.7 ± 3.6% to 13.1 ± 1.8% in the liraglutide group, from 33.0 ± 3.5% to 19.6 ± 2.1% in the gliclazide group, and from 35.1 ± 2.3% to 18.4 ± 2.2% in the metformin group (P < 0.001 for all treatment groups, final vs. baseline)

-p

Reduction in liver fat following liraglutide treatment was greater than that following gliclazide treatment (P = 0.001)

re

Mean ALT 49 IU/mL, AST 31 IU/L

Jo u

rn al P

Both liraglutide and metformin treatments significantly reduced weight and improved liver function tests

Frossing, 2017 (28) Denmark, LIPT-study (Fair)

Obese non-diabetic women with polycystic ovary syndrome and NAFLD (assessed by MR spectroscopy) Age: 47 y

Sex: 100% female BMI, kg/m2: 33 Diabetes: NR Mean ALT and AST: NR

A. Placebo (n = 24) B. Liraglutide mg/d (n = 48)

1.8

Duration: 26 weeks

HbA1c levels were lower in the liraglutideand metformin-treated groups than in the gliclazidetreated group Compared with placebo, liraglutide treatment reduced body weight by 5.2 kg (-5.6% from baseline), intrahepatic triglyceride content (as measured by MR spectroscopy) by 44% and the prevalence of NAFLD by about two-thirds (all P < 0.01)

Nausea (liraglutide 79%; placebo 13%; P < 0.01) and constipation (liraglutide 26%; placebo 0%; P < 0.01) were the most frequent AEs

Liraglutide treatment caused significant reductions in fasting plasma glucose (liraglutide vs placebo, mean between-group difference [95% CI], −0.24 [−0.44 to −0.04] mmol/L; mean

29

Page 29 of 38

Yan, 2019 (29) China (Fair)

Patients with NAFLD (assessed by MRIPDFF) and type 2 diabetes who were unable to maintain good glycaemic control with metformin Age: 44 y Sex: 69% male

A. Liraglutide mg/d (n = 24)

1.8

B. Insulin glargine 0.2 IU/kg/d (n = 24) C. Sitagliptin mg/d (n = 27)

100

HbA1c [95% CI], −1.38 [−2.48 to −0.28] mmol/mol) In the liraglutide and sitagliptin groups, hepatic fat content, measured by MRI-PDFF. significantly decreased from baseline to week 26 (liraglutide, 15.4±5.6% to 12.5±6.4%, P < 0.001; and sitagliptin, 15.5 ± 5.6% to 11.7 ± 5.0%, P = 0.001)

Duration: 26 weeks Although this change was greater with liraglutide than sitagliptin, it was not significantly different between the two groups (−4.0 vs. −3.8%; P = 0.91)

BMI, kg/m2: 29.8

of

HbA1c: 7.7%

Six patients in the liraglutide group withdrew from the study (four lost to follow-up, one for protocol violations, and one for AEs), one patient in the sitagliptin group was lost to follow-up, and three patients in the insulin glargine group withdrew for protocol violations

Mean ALT 43 IU/mL, AST 33 IU/mL

-p

ro

In contrast, hepatic fat content did not change significantly from baseline in the insulin glargine group

Jo u

rn al P

re

HbA1c improved significantly in all treatment groups (liraglutide, 7.8 ± 1.4% to 6.8 ± 1.7%, P < 0.001; sitagliptin, 7.6 ± 0.9% to 6.6 ± 1.1%, P = 0.016; and insulin glargine, 7.7 ± 0.9% to 6.9% ± 1.1%, P = 0.013). However, ΔHbA1c did not differ across treatment groups In the liraglutide and sitagliptin groups (but not in the insulin glargine groups), significant decreases in body weight and BMI were observed

DPP-4 inhibitors

Macauley, 2015 (30) UK (Fair)

Patients with type 2 diabetes and NAFLD (assessed by MRIPDFF) on stable metformin therapy Age: 61.4 y Sex: 64% male BMI, kg/m2: 30.3

A. Vildagliptin 50 mg bid (n = 22) B. Placebo (n = 22) Duration: 6 months

Body weight decreased by 1.6 ± 0.5 vs. 0.4 ± 0.5 kg in the vildagliptin and placebo groups, respectively (P = 0.08)

There were no meaningful differences between the vildagliptin and placebo groups in the overall AEs

Mean hepatic fat content (assessed by MRI-PDFF) decreased significantly with vildagliptin treatment from 7.3 ± 1.0% at baseline to 5.3 ± 0.9% at endpoint (P = 0.001)

HbA1c 6.4%

30

Page 30 of 38

Cui, 2016 (31) United States (Fair)

Patients with NAFLD (assessed by MRIPDFF) and with prediabetes (n = 25) or controlled type 2 diabetes (n = 25)

No change in the placebo group (5.4 ± 0.7% to 5.4 ± 1.0%; P = 0.48). The between-group difference in change from baseline was significant (P < 0.013)

A. Sitagliptin 100 mg (n = 25) B. Placebo (n = 25) Duration: 24 weeks

Compared to baseline, there were no significant differences in end-oftreatment MRI-PDFF for sitagliptin (18.1% to 16.9%, P = 0.27) or placebo (16.6% to 14.0%, P = 0.07)

-p

Sex: 62% male BMI, kg/m2: 31.8

(32)

rn al P

2017

Patients with uncomplicated type 2 diabetes and NAFLD diagnosed by ultrasound

Jo u

Deng, China (Fair)

re

HbA1c 6.2% Mean ALT 42 IU/mL; AST 28 IU/mL

Age: 64 y

NR

ro

Age: 54 y

Mean plasma ALT fell from 27.2 ± 2.8 to 20.3 ± 1.4 IU/L in the vildagliptin group (P = 0.001), and did not change in the placebo group (29.6 ± 3 to 29.6 ± 3.7 IU/L; P = 0.44) Sitagliptin was not significantly better than placebo in reducing liver fat content measured by MRI-PDFF (mean difference between sitagliptin and placebo arms: -1.3%, P = 0.40)

of

Mean ALT 28 IU/mL

A. Sitagliptin 50 mg to 100 mg (n = 36) B. Diet and exercise (n = 36)

No significant differences for changes in serum transaminase levels, HOMA-IR or MRE-derived liver stiffness between the two groups No differences for changes in serum AST (P = 0.99) and ALT (P = 0.97) between treatment with sitagliptin vs. diet and exercise

Duration: 52 weeks

Greater decrease in HbA1c with sitagliptin (-0.81% from baseline) vs. diet and exercise (-0.25%), P < 0.01 at 52 weeks

A. Liraglutide mg/d (n = 24)

In the liraglutide and sitagliptin groups, liver fat content, measured by MRIPDFF. significantly decreased from baseline to week 26 (liraglutide, 15.4 ± 5.6% to 12.5 ± 6.4%, P < 0.001; and sitagliptin, 15.5

Sex: 75% male BMI, kg/m2: 24

NR (70 patients completed the trial)

Diabetes: HbA1c 7.4%

Yan, 2019 (29) China (Fair)

Mean ALT 35 IU/mL, AST 32 IU/mL Patients with NAFLD (assessed by MRIPDFF) and type 2 diabetes who were unable to maintain glycaemic control with metformin

1.8

B. Insulin glargine 0.2 IU/kg/d (n = 24)

Six patients in the liraglutide group withdrew from the study (four lost to follow-up, one for protocol violations, and one for AEs), one patient in the sitagliptin group was

31

Page 31 of 38

Age: 44 y

C. Sitagliptin mg/d (n = 27)

100

± 5.6% to 11.7 ± 5%, P = 0.001)

Sex: 69% male Duration: 26 weeks BMI, kg/m2: 29.8 HbA1c: 7.7% Mean ALT 43 IU/mL, AST 33 IU/mL

Although this change was greater with liraglutide than sitagliptin, it was not significantly different between the two groups (−4.0 vs. −3.8; P = 0.91)

lost to follow-up, and three patients in the insulin glargine group withdrew for protocol violations

Liver fat content did not change significantly from baseline in the insulin glargine group

-p

ro

of

HbA1c improved significantly in all treatment groups (liraglutide, 7.8 ± 1.4% to 6.8 ± 1.7%, P < 0.001; sitagliptin, 7.6 ± 0.9% to 6.6 ± 1.1%, P = 0.016; and insulin glargine, 7.7 ± 0.9% to 6.9 ± 1.1%, P = 0.013). However, ΔHbA1c did not significantly differ across treatment groups

Jo u

rn al P

re

In the liraglutide and sitagliptin groups, significant decreases in body weight were observed Abbreviations: AE, adverse effects; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; MR, magnetic resonance; MRI-PDFF, magnetic resonance imaging-proton density fat fraction; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; NR, not reported.

32

Page 32 of 38

Table IV. Placebo-controlled or active-controlled RCTs of SGLT2 inhibitors for treatment of NAFLD (ordered by publication year). Interventions (group sizes), Duration

Efficacy/effectiveness outcomes A vs. B (vs. C)

Adverse effects

Patients with NAFLD (assessed by MRIPDFF) and type 2 diabetes inadequately controlled on metformin

A. Placebo (n = 91)

At week 24, placebocorrected changes with dapagliflozin were as follows: body weight, -2.08 kg [95% confidence interval (CI)= -2.84 to -1.31; P < 0.0001]; waist circumference, -1.52 cm (95% CI= -2.74 to -0.31; P = 0.014); total fat mass, -1.48 kg (95% CI = -2.22 to -0.74; P < 0.0001)

In B vs. A group: serious AEs were reported in 6.6% vs. 1.1%, respectively; events suggestive of vulvovaginitis, balanitis, and related genital infection in 3.3 vs. 0%; and lower urinary tract infections in 6.6 vs. 2.2%

Age: 58.6 y

B. Dapaglifozin 10 mg/d (n = 91) Duration: 24 weeks (with a 78-wk siteand patient-blinded extension period)

Sex: 46.7% male Ethnicity: 100% White BMI, kg/m2: 31.9 HbA1c 7.2%

In a subset of patients (n = 42 in the placebo arm and n = 38 in the active drug arm), MRI-PDFF was also performed

In the MR sub-study: Dapagliflozin produced significantly greater mean reductions from baseline in visceral adipose tissue compared with placebo at 24-week

-p

Mean ALT and AST: NR

of

Population, Demographics

ro

Investigator, Year, Country, Trial name (Quality rating) Bolinder, 2012 (33) International (Fair)

(34), Japan

Patients with poorly controlled type 2 diabetes and NAFLD on ultrasound

A. Pioglitazone 15-30 mg/d (n = 34) B. Ipraglifozin mg/d (n = 32)

50

Age: 58 y

Jo u

Ito, 2017 Multicenter (Fair)

rn al P

re

Change from baseline at 24-week in mean percent hepatic fat content with dapagliflozin was -2.35% and -1.53% with placebo, resulting in a not significant placebocorrected difference of 0.82% (95% CI = -2.97 to 1.33; P = 0.45) Mean liver-to-spleen attenuation ratio on computed tomography at week 24 increased by 0.22 (from 0.80 ± 0.24 to 1.0 ± 0.18) in the ipragliflozin group and 0.21 (from 0.78 ± 0.26 to 0.98 ± 0.16) in the pioglitazone group (P = 0.90)

Sex: 48.5% male BMI, kg/m2: 30 HbA1c 8.4%

Mean ALT 55 IU/mL, AST 41 IU/mL

Duration: 24 weeks

Serious AEs: NR

AST, ALT and GGT levels, HbA1c and HOMA-IR were similarly reduced in the two treatment groups FIB4 score was similarly reduced in the two treatment groups Body weight and visceral fat area showed significant reductions only in the ipragliflozin group

33

Page 33 of 38

Cusi, 2018 (35) USA (Fair)

Patients with type 2 diabetes who were unable to maintain glycaemic control (most patients had NAFLD on MRI-PDFF) Age: 58 y

A. Placebo (n = 30) B. Canaglifozin 100 mg/d titrated up to 300 mg/d (n = 26) Duration: 24 weeks

compared with the pioglitazone group A not significant larger absolute decrease in hepatic fat content occurred with canagliflozin (hepatic fat content: −4.6% [−6.4; −2.7]) vs. placebo (−2.4% [−4.2; −0.6]; P = 0.09)

Serious AEs: NR SAE: 3% vs. 4%

In patients with NAFLD, the decrease in hepatic fat content was −6.9% (−9.5; −4.2) vs. −3.8% (−6.3; −1.3; P = 0.05), and strongly associated with the magnitude of weight loss

Sex: 66% male Ethnicity: 67% White BMI, kg/m2: 32

of

HbA1c 7.7%

Body weight loss ≥5% with a ≥30% relative reduction in hepatic fat content occurred more often with canagliflozin (38% vs. 7%, P = 0.009)

ro

Mean ALT 30 IU/mL, AST 25 IU/mL

-p

Mean hepatic fat content measured by magnetic resonance spectroscopy: 12.5%

Patients with type 2 diabetes and NAFLD (assessed by MRIPDFF)

A. Placebo (n = 25)

Age: 50 y

Duration: 20 weeks

Jo u

Kuchay, 2018 (36) India, E-LIFT (Fair)

rn al P

re

Canagliflozin reduced HbA1c (placebosubtracted change: −0.71% [−1.08; −0.33]) and body weight (−3.4% [−5.4; −1.4]; both P < 0.001)

Sex: 60% male BMI, kg/m2: 29.5 HbA1c 9% Mean ALT 64 IU/L, AST 44 IU/L

B. Empaglifozin 10 mg/d (n = 25)

Hepatic insulin sensitivity improved with canagliflozin (P < 0.01), but not muscle or adipose tissue insulin sensitivity (measured by euglycaemic insulin clamp) Empagliflozin was significantly better at reducing liver fat content (mean MRI-PDFF difference between the empagliflozin and control groups 24.0%; P < 0.0001) Compared with baseline, significant reduction was found in the end-oftreatment MRI-PDFF for the empagliflozin group (16.2% to 11.3%; P < 0.0001) and a nonsignificant change was found in the control group (16.4% to 15.5%; P = 0.057)

Serious AEs: NR In the empagliflozin arm, 22 patients completed the study, with 3 developing AEs related to the study medication In the control arm, 20 patients completed the study, with three lost to follow-up and two patients discontinuing because of work schedule conflicts

The two groups showed significant differences for change in serum ALT level (P = 0.005) and non-

34

Page 34 of 38

Eriksson, 2018 (37) Multicentre, Sweden, EFFECT-II trial (Fair)

Patients with type 2 diabetes and NAFLD (assessed by MRIPDFF)

A: Placebo (n = 21) B: Omega-3 carboxylic acids (OM3CA) 4 g/d (n = 20)

Age: 65.5 y

BMI, kg/m2: 31.2 HbA1c 7.5% Mean ALT and AST: NR

C: Dapaglifozin 10 mg/d (n = 21) D: Omega-3 carboxylic acids 4 g/d + dapaglifozin 10 mg/d (n = 22)

Only the combination treatment reduced liver fat content (P = 0.046) and total liver fat volume (relative change, −24%, P = 0.037) in comparison with placebo

Duration: 12 weeks

All active treatment groups had similar total percentages of AE reporting (70.0– 77.3%), which were higher than in the placebo group (47.6%) More participants reported AEs when using dapagliflozin and OM-3CA (n = 15, 68.2%) than when using dapagliflozin monotherapy (n= 7, 33.3%), OM-3CA monotherapy (n = 8, 40%) or placebo (n= 6, 28.6%)

of

Sex: 70% male

significant differences for AST (P = 0.212) and GGT (P = 0.057) levels All active treatments significantly reduced hepatic fat content from baseline, relative changes: OM-3CA, −15%; dapagliflozin, −13%; OM3CA + dapagliflozin, −21%

-p

ro

Dapagliflozin monotherapy, but not the combination with OM-3CA, reduced serum transaminase and GGT levels

Age: 56 y Sex

re

Patients with type 2 diabetes and NAFLD (assessed by Fibroscan and CAP measurement)

A: Placebo (n = 24)

rn al P

Shimitzu, 2019 (38), Japan (Fair)

Jo u

BMI, kg/m2: 28.0 HbA1c: 7.8%

Mean ALT 36 IU/L, AST 27 IU/L

Dapagliflozin alone and in combination with OM-3CA improved HbA1c and reduced body weight In week 24, there was a significant decrease in controlled attenuation parameter (CAP) from 314 ± 61 to 290 ± 73 dB/m (P = 0.042) in the dapagliflozin group, but not in the control group

B: Dapagliflozin mg/day (n = 33) Duration: 24 weeks

5

NR

Liver stiffness measurement (LSM) tended to decrease from 9.49 ± 6.1 to 8.01 ± 5.8 kPa in the dapagliflozin group. In 14 patients from this group with LSM values ≥ 8.0 kPa, LSM decreased significantly from 14.7 ± 5.7 to 11.0 ± 7.3 kPa (P = 0.016) Serum ALT and GGT levels decreased significantly in the dapagliflozin group, but not in the control group Changes in BMI and HbA1c in the control vs. the dapagliflozin groups were 0.0 (−0.55, 0.50) vs. −0.8 (−1.25, −0.07) kg/m2; and HbA1c − 0.3 (−0.5, 0.5) vs.

35

Page 35 of 38

Kahl, 2019 (39), Germany (Fair)

Patients with wellcontrolled type 2 diabetes and NAFLD (assessed by MR spectroscopy) Age: 62 y

A: Placebo (n = 42) B: Empaglifozin 25 mg/day (n = 42) Duration: 24 weeks

Sex: 69% male Ethnicity: 100% White

Serious AEs: NR All treatment groups had similar total percentages of AEs (5 events in EMPA and 7 events in placebo groups)

Weight loss occurred only with EMPA (placebocorrected change -2.5 kg [3.7, -1.4 kg]; P < 0.001), while no placebocorrected change in tissuespecific insulin sensitivity was observed

BMI, kg/m2: 32.2

of

HbA1c 6.6%

ro

Mean ALT 35 IU/mL, AST 25 IU/mL

-p

EMPA treatment also led to placebo-corrected changes in uric acid (-74 mmol/L [-108,-42 mmol/L]; P < 0.001) and highmolecular-weight adiponectin (36% [16, 60%]; P < 0.001) levels from 0 to 24 weeks

re

rn al P

Mean hepatic fat content measured by magnetic resonance spectroscopy: 13% (80% of patients had hepatic steatosis at baseline)

−0.8 (−1.3, −0.5)%, respectively EMPA treatment resulted in a placebo-corrected absolute of 21.8% (95% CI 23.4, 20.2%; P = 0.02) and relative change in liver fat content of -22% (-36, -7%; P = 0.009) from baseline to end of treatment, corresponding to a 2.3-fold greater reduction

Serum ALT and GGT levels were reduced with similar effect sizes in EMPA and placebo after 24 weeks

Jo u

Abbreviations: AE, adverse effects; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; MR, magnetic resonance; MRI-PDFF, magnetic resonance imaging-proton density fat fraction; NAFLD, non-alcoholic fatty liver disease; NASH, non-alcoholic steatohepatitis; NR, not reported.

36

Page 36 of 38

Supplementary Table S1. Randomized controlled trials excluded at the stage of eligibility according to the PRISMA flow diagram.

Joy et al. (4)

Metformin therapy combined with dietary treatment was effective for the reduction of hepatic steatosis and fibrosis evaluated by liver ultrasonography and elastography, respectively Pioglitazone reduced liver fibrosis in patients with type 2 diabetes rather than in patients with pre-diabetes

Unsatisfactory study design (low quality)

Sitagliptin did not improve fibrosis score or NAS after 24 weeks when compared to placebo Lixisenatide increased the proportion of obese or overweight patients with type 2 diabetes who achieved normalisation of serum ALT Metformin had a small beneficial effect on serum liver enzymes or liver histology in patients with NASH Metformin decreased serum aminotransferase levels compared to dietary alone. No significant differences in necro-inflammatory activity or fibrosis were observed between the groups Liraglutide 1.8 mg/day reduced serum ALT levels compared to placebo (and was dose responsive); lower doses of liraglutide (0.6 and 1.2 mg/day) had similar effects as placebo. In LEAD2, liraglutide improved liver fat content in a dose dependent way; however, there was no significant difference between liraglutide and placebo after adjusting for weight loss or HbA1c Canagliflozin provided significant improvements in ALT, AST and GGT levels vs.

Unsatisfactory sample size

Shields et al. (6)

19 patients with NASH were randomized to dietary or metformin for 12 months

36 patients NASH were randomized to dietary treatment alone or to diet plus metformin therapy for 6 months

Jo u

Uygun et al. (7)

rn al P

re

Gluud et al. (5)

101 NASH patients with type 2 diabetes or pre-diabetes were randomly assigned to pioglitazone or placebo for 18 months 12 patients with biopsyproven NASH were randomized to sitagliptin or placebo for 24 weeks Individual patient data of 12 RCTs on lixisenatide vs. placebo and 3 RCTs with active comparators

Exclusion reasons Unsatisfactory study design (low quality)

of

Bril et al. (3)

Main findings Silymarin, pioglitazone and vitamin E improved serum aminotransferases in patients with NAFLD

ro

Handzlil et al. (2)

Study characteristics 150 patients with NAFLD were randomized to five groups: lifestyle plus placebo, metformin, silymarin, pioglitazone, or vitamin E for 3 months 42 patients with NAFLD randomized to dietary treatment alone (n = 21) or to diet plus metformin (n = 21)

-p

Investigator, Ref. Anushiravani et al. (1)

Armstrong et al. (8)

LEAD & LEAD-2 programs: 3258 patients with type 2 diabetes who were unable to maintain good glycaemic control with diet and exercise alone, or with oral hypo-glycaemic treatment

Leiter et al. (9)

Patients with type 2 diabetes pooled from four placebo-controlled trials with canaglifozin 100 and

Data already included in another eligible RCT (see ref. #23 of the manuscript)

Unsatisfactory study design (systematic review with individual patient data meta-analysis)

Unsatisfactory sample size

Unsatisfactory sample size

Unsatisfactory inclusion criteria (no information on alcohol intake or diagnosis of NAFLD by imaging or biopsy)

Unsatisfactory inclusion criteria (no information on alcohol intake or diagnosis

37

Page 37 of 38

of NAFLD by imaging or biopsy)

Empaglifozin 10-25 mg/day provided significant improvements in serum ALT, levels vs. either placebo or glimepiride treatment. ALT reductions were largely independent of concomitant changes in body weight or HbA1 More patients on combination therapy (pioglitazone plus vitamin E) improved liver histology versus placebo, but not with vitamin E alone

Unsatisfactory inclusion criteria (no information on alcohol intake or diagnosis of NAFLD by imaging or biopsy)

4. 5.

6.

7. 8.

9. 10.

11.

re

3.

rn al P

2.

Anushiravani A, Haddadi N, Pourfarmanbar M, Mohammadkarimi V. Treatment options for nonalcoholic fatty liver disease: a double-blinded randomized placebo-controlled trial. Eur J Gastroenterol Hepatol 2019;31:613-7. Handzlik G, Holecki M, Kozaczka J, Kukla M, Wyskida K, Kędzierski L, et al. Evaluation of metformin therapy using controlled attenuation parameter and transient elastography in patients with non-alcoholic fatty liver disease. Pharmacol Rep 2019;71:183-8. Bril F, Kalavalapalli S, Clark VC, Lomonaco R, Soldevila-Pico C, Liu IC, et al. Response to pioglitazone in patients with nonalcoholic steatohepatitis with vs without type 2 diabetes. Clin Gastroenterol Hepatol 2018;16:558-66. Joy TR, McKenzie CA, Tirona RG, Summers K, Seney S, Chakrabarti S, et al. Sitagliptin in patients with non-alcoholic steatohepatitis: A randomized, placebo-controlled trial. World J Gastroenterol 2017;23:141-50. Gluud LL, Knop FK, Vilsbøll T. Effects of lixisenatide on elevated liver transaminases: systematic review with individual patient data meta-analysis of randomised controlled trials on patients with type 2 diabetes. BMJ Open 2014;4:e005325. Shields WW, Thompson KE, Grice GA, Harrison SA, Coyle WJ. The effect of metformin and standard therapy versus standard therapy alone in nondiabetic patients with insulin resistance and nonalcoholic steatohepatitis (NASH): a pilot trial. Therap Adv Gastroenterol 2009;2:157-63. Uygun A, Kadayifci A, Isik AT, Ozgurtas T, Deveci S, Tuzun A, et al. Metformin in the treatment of patients with nonalcoholic steatohepatitis. Aliment Pharmacol Ther 2004;19:537-44. Amstrong MJ, Houlihan DD, Rowe IA, Clausen WH, Elbrond B, Gough SC, et al. Safety and efficacy of liraglutide in patients with type 2 diabetes and elevated liver enzymes: individual patient data meta-analysis of the LEAD program. Aliment Pharmacol Ther 2013;37:234-42. Leiter LA, Forst T, Polidori D, Balis DA, Xie J, Sha S. Effect of canagliflozin on liver function tests in patients with type 2 diabetes. Diabetes Metab 2016;42:25-32. Sattar N, Fitchett D, Hantel S, George JT, Zinman B. Empagliflozin is associated with improvements in liver enzymes potentially consistent with reductions in liver fat: results from randomised trials including the EMPA-REG OUTCOME¬Æ trial. Diabetologia 2018;61:2155-63.

Jo u

1.

-p

REFERENCES for the table

Data already included in another eligible RCT (see ref. #23 of the manuscript)

of

Bril et al. (11)

either placebo or 100-mg sitagliptin treatment

ro

Sattar et al. (10)

300 mg/day (n=2313), and two active-controlled trials of canagliflozin 300 mg/day vs. sitagliptin 100 mg/day (n=1488) Patients with type 2 diabetes included in the EMPA-REG OUTCOME trial (n=7020), pooled data from four 24-week placebocontrolled trials (n=2477) and a trial of empagliflozin vs. glimepiride over 104 weeks (n=1545) Patients with type 2 diabetes and biopsy-proven NASH (n=105) were randomized to vitamin E 400 IU b.i.d., vitamin E 400 IU b.i.d. plus pioglitazone 45 mg/day, or placebo. Eightysix patients completed the 18-month study

Bril F, Biernacki DM, Kalavalapalli S, Lomonaco R, Subbarayan SK, Lai J, et al. Role of vitamin E for nonalcoholic steatohepatitis in patients with type 2 diabetes: a randomized controlled trial. Diabetes Care 2019;42:1481-8.

38

Page 38 of 38